Meta Connect 2026: Muse Opens a New Battle for Brand Discovery

Mark Zuckerberg did not say the word "advertising" once during his Meta Connect 2026 keynote.

This is a remarkable detail from the CEO of the world's second-biggest ad company.

The headline product was Muse, Meta's new personal AI agent, which is already live on the web, iOS, Android, and WhatsApp in the U.S.

Muse is free at the entry tier, with Power and Maximum subscriptions priced at $20 and $100 a month for heavier use.

It connects to email, calendars, payments, and services including Gmail, Spotify, Ticketmaster, Shopify, and OpenTable, and will soon come to Meta's AI glasses.

Once connected, Muse can send emails, book travel, fill out forms, compare products, and complete a purchase using Link by Stripe for checkout.

Meta says the agent will earn through transaction activity, not advertising, though its confirmed monetization model so far is the two-tier offering.

Amazon then blocked Muse from shopping on its site, escalating a dispute over whether Meta's agent could access Amazon's marketplace.

The conflict highlights the growing role AI agents could play in shaping how consumers discover products.

Muse could make this decision before a shopper reaches a product page, giving brands a new discovery layer to prepare, bringing AI agents into the buying journey.

Two Companies Fighting Over the Same Customer

Amazon's block came with a popup warning that "continued access by an unauthorized AI agent violates Amazon's Conditions of Use."

It told the press it had asked Meta to remove Amazon from Muse's shopping experience before the block went up, and that Meta declined.

The dispute reads as a contract fight on the surface, but underneath, it is a fight over who controls product discovery.

If Muse decides what a shopper sees, compares, and buys, Meta's agent is making the call that Amazon's own search and advertising system makes.

Amazon already runs its own agent, Alexa for Shopping, which has been built to keep this same decision inside Amazon's own walls.

Letting a Meta-built agent make purchasing decisions on Amazon's marketplace would give a competitor the exact layer Amazon has already built for itself.

Amazon's Conditions of Use, last updated in August, require any shopping agent to identify itself openly in every request and avoid mimicking human browsing behavior.

Whether Muse met this bar is Amazon's allegation, and Meta disputes it.

Either way, both companies are treating the dispute as a fight over territory that each tech giant wants to own.

Muse Changes the Path to Purchase

Social advertising has run on one assumption for at least a decade.

A brand pays to interrupt a scroll, a person sees the ad, and then the person decides whether to click through to a product page.

But Muse changes what happens between the interruption and the purchase.

The agent evaluates preferences, compares products across services, and can complete the transaction without the person ever visiting a product page.

However, winning in Meta's ad auction does not mean a brand will be visible to Muse.

Muse reads structured data such as titles, specifications, and reviews, along with whatever a service exposes about a product through its connectors.

A brand with strong ad creative and thin, unstructured product data gives Muse very little to evaluate, compare, or recommend.

This means that structured product content will become increasingly important as brands prepare for agent-led discovery.

The Launch's Trust Problem

Muse arrived two weeks after Meta agreed to an $18 billion multistate settlement over social media's harm to children.

Muse asks users to connect their email, payment methods, and calendars directly to Meta, a level of access the company has never asked consumers for before.

TechCrunch coverage also raised Meta's history with the 2018 Cambridge Analytica scandal and a $5 billion FTC settlement the same year as reasons consumer trust here cannot be assumed.

Meta has published technical documentation describing Muse's security architecture.

It includes a claim that the agent cannot see a user's passwords or stored payment details.

Nobody knows yet whether people will actually trust Meta enough to hand over this access at scale, and every brand watching this launch has money tied to how this plays out.

If people don't trust Muse with their accounts, there's nothing here for a brand to optimize toward.

If Muse does catch on, it becomes the new front door to customers that brands have spent a decade reaching through ads.

Getting Ready for an Agent-Led Discovery Layer

Morgan Stanley has estimated Muse could generate $1.3 billion in annual revenue by 2028, built on transaction volume tied to completed purchases.

This kind of money moves quickly, which means brands need to start planning for it now.

Amazon's response confirms the same timeline, as a company does not block a rival's agent within two weeks of launch over something it expects to matter years from now.

Retail media budgets are typically set a year in advance, while a new discovery layer like Muse is already creating decisions those budgets did not account for.

Other analysts see even larger long-term numbers attached to Muse, with Raymond James projecting more than $50 billion in eventual revenue as the agent scales.

Three things belong on this plan now:

• Map the agent shopping journey: Identify every step from product discovery to checkout and assign ownership across marketing, ecommerce, and product teams.
• Prepare agent-specific measurement: Define how referrals, recommendations, and purchases from AI agents will be tracked in existing analytics and attribution systems.
• Review agent access and permissions: Document which AI agents can access product, pricing, inventory, and checkout data across each commerce platform.

Each of these three takes lead time to build, which is why they belong on a calendar today, ahead of the next major Muse update.

Our Take: Is This the Beginning of the End for Social Ads?

This could mark the beginning of a second channel, which will run in parallel to the one brands already know.

We don't think Zuckerberg leaving "advertising" out of his own keynote was an accident.

Meta is building a business model that earns from completed purchases, and it needs Muse to be useful enough that people route their decisions through it.

If this happens at scale, brand visibility inside Muse will become its own discipline that runs alongside ad performance as a second measurement.

Amazon's choice to block Muse within two weeks of its launch also confirms how seriously at least one competitor is treating this possibility.

Preparing product data now, ahead of wide consumer adoption, puts a brand ahead of competitors still treating Muse as a launch-week curiosity.

Waiting for the shape of this to become obvious means playing catch-up with an agent that already has years of purchase history on faster-moving competitors.

Brands preparing for agent-led shopping need agencies who understand how AI discovery is changing what earns visibility outside a paid ad.

Check out the agentic AI companies in our directory.